Portal de Periódicos da CAPES

Indoor Domain Model for Dialogue Systems

2009; Springer Science+Business Media; Linguagem: Inglês

10.1007/978-3-642-02713-0_54

1611-3349

Porfírio P. Filipe, Nuno Mamede,

Context-Aware Activity Recognition Systems

The design of spoken language applications that allow people to talk with machines/computers, in the same way that they talk with each other, is materialized as a Spoken Dialogue System (SDS). This paper presents a knowledge modeling approach to allow spontaneous configuration of SDS. Our approach focus on the representation and management of the domain knowledge that is aggregated at runtime and aims to update the dialogue management strategy. To do so, one developed an autonomous Environment Interaction Manager (EIM). When working on the indoor environment, the domain knowledge reflects the plan of the building and the SDS controllable resources. The building is modeled as a dynamic aggregation “part-whole” of controllable resources. Each resource owns and shares a semantic interface that makes available its task set manipulated by the SDS. These ideas have been applied with success in our lab, modeling the semantic interface aggregation under the semantic web vision.